You are on page 1of 6
A Simple Means for Reducing the Risk of Progressive Collapse Debond evelopment of catenary reinforcing bars are shown to enhance d ction in beams by Yihai Bao, H.8, Lew. Fahim Sadek, and Joseph Main reinforwed concrete (RC) fiame structure, comprising two beams and three columns, have shown that the development of catenary action under a central column loss scenario is limited by fracture of the bottom flexural reinforcement in the beams.’ Similar behavior has been observed in other tests of RC frames or assemblies under column removal scenarios. Most of the tested specimens failed in a similar manner: tensile fracture of the reinfor- wg bars at the beam ends. Figure 1 shows one of the fractured bottom bars at a wid Rexural crack that devel oped at the beam-central column interface during a subframe assembly test. Strain localization occurred in the exposed bar segment at the crack due to the greater axial constraint provided by concrete bonded to the embedded bar segments on each side of the crack. Validated, detailed finite element (FE) models ofan intermediate moment frame (IMF) assembly (complying with Section 21.12 of ACI 318402%) have confirmed the existence of this strain localization. As shown in Fig, 2, the FE models verify thava bonded bortom reinforcing bar exhibits a sharp cae in stain near an unsupported central column, while a bottom bar with a debonded length of one beam depth exhibits minimal stain localization. This indicates that «lebonding the boitom reinforcing bars atthe beam ends will delay fiacture,enabling a beam to sustain larger rotations, sased catenary forces, and cal load-carrying capacity under acolumn removal scenario. Figure 3 schematically illustrates the concept of using a debonding technique to avoid strain localization at domin- ant crack openings. Based on this concept.a simple ap. proach fordebonding. bars is proposed herein,and this approach is validated through an experimental stud Furthermore, its effectiveness in resisting disproportionate collapse (commonly known as progressive collapse) is R=: tests of planar subassemblies of a 10story verified by computational analysis. The optimum debonded length and the behavior under seismie loading cond: are also evaluated numerically In general, the negative bending moment capacity of beam end sections is larger than the positive benkling moment capacity. Under a column loss scenario, the bortom reinforcing bars are thus likely to fracture before the top reinforcing bars. On that basis, as well as the face that debonding of the top bars could have undesirable effects on the connection stiffness, only bottom bar debonding was considered in this study Experimental Investigation Three specimens were investigated under uniavial tensile load to demonstrate the effectiveness of a technique for debonding reinforcing bars feom the surrounding concrete, The efficacy of debonding reinforcing bars at a cracked zone was aso evaluated numerically to determine its eilfectiveness in delaying the fracture of bars and providing, enhanced ductile behavior, Fig, |: Observed fracture of botlom reinforcing bar during beam-column assembly test nerete international DECEMBER 2013 33, Specimen design Fach specimen consisted of wo 10 x48 ‘concrete cylinders connected by one No.8 (No. . bar (Fig. 4). 1/4 in. (644 mm) gap beoween the two cylinders ‘was designed to represent a crack opening. The specimens were pulled in the longitudinal direction until the winforcing bar fiactured. The force was applied through a Tshaped steel loading fisture ateach end of specimen. The force was transmitte! from the fixture to the concrete by means of four high-strength 7/8 in. (22.2 mm) diameter threaded barswhich \wereconnected toa 2 in,(51 mm) thick circular stel disk that send asthe flange of the Fsaction of the loading fixture. The concrete fommvork comprised round paperboard tubes. The reinforcing bars were debonded using polyolefin beatshrink tubing (when heated, the tubing shrinks in the radial direction to fi tightly over the bar [Fig. 5 mens A,B,and C were fabricated with 0,2, and and 203 mm) debonded lengths on each side of the gap between the concrete cylinders, respectively Material properties The specimens were fabricated with self-consolidating concrete having a nominal compressive strength of 6000 psi Distance from the face of the central calumn, mm o ims 2323088 4058 S08 02s. ‘$odebonding + debondedlenath of 1.00 ous. (41.4 MPa), Atthe time of testing, the compressive strength 000 psi (93 MPa) based on the average of three 6 x 12 in (152 x 305 mm) cylinder breaks. The No. 8 (No. 25) reinforcing bars met ASTM A706 GR 60 yents, with measured yield and ultimate strengths ‘oF 70,000 psi (483 MPa) and 103,000 psi (710 MPa), respectively. The yield strain was 0.26596 and the fracture strain was 169 with a gauge length of 8 in. (203 mm) ‘of the concrete was 1 Instumentation and test setup Strain in the reinforcing bar was measured using strain gauges bonded to the bar surface. Gauges were applied to the bar over an 8 in. (203 mm) length on each side of the gap. The relative movement between two cross sections located at 3 in, (76 mm) from the gap was measured by displacement transducers (Pig. 4). Tensile loadin applicdl uncler displacement control at a rate of 15 ing minute (3.8 mm/minute). Experimental results The specimens were loaded continuously until failure ‘occurred. For all chive specimens, failure was characterized by fracture of the reinforcing bar within the 1/4 in. (6.4 mm) gap. As shown in Fig. 6, the specimen with no debonding (Specimen A) developed a cone-shaped concrew spall atthe five surface of the crack opening, In contrast, the free surface of the crack opening exhibited no signifi cantconcrete damage in specimens with debonding i 3 () Steel loading fixture ASTM A706 No. 8 bar ode incdmeter head _/ (hin. dlameter Sal x ¢ ¢ z 7 [ uhend G/Sin-ciameteriy i af fl e@ fon ow Distance from the face ofthe entra olurm, in Vee #2, _[thin) section a cin pe a em cigs Een LE pe fpospil lpia lest homo {Seuney One neteacepns rerotea ren ot Celactan chest masekep bond ten terrsaneSiorno evendcensobare prided nBserees e Debonded Zone Fig. 3: Schomatic representation of the effect of debonding on the stan distibution in a reinforcing bar af and near a crack, 34 BER 203 Conerete international Fig. 4: Test specimen and apparatus: (a) schematic of specimen _goomety; (b) schematic of fest setup; and (c) photo of fost (Specimen B and C). Figure 7 shows the applied load as a function of rel wage length for the thive specimens. While the peak lead almost identical for the three specimens, significant differences are observed among the peak displacements of the three specimens prior to failure. The specimen without debonding (Specimen A) failed at a displacement of about 1LAS in. (37 mm). Specimen B, with a debonded length of 2in, (51 mm), failed at a smaller displacement of approximately 0.9 in. (23 mm), while Specimen C, with a debonded length of 8 in, (203 mm), failed aca larger dlisplacement of approximately 2 in, (51 mm). ‘These results show thardebonding over very short length «can actually result in eadicr Fracture than having no deborxling Thiscan be explained by the defacto debonding, (oss of bord) that occumre within the cone-shaped spall of Specimen A. Is, apparent that ifthe debonded length is kess chan the depth of this damage zone in the concrete, then strains inthe reinforcing, bar will be localized within che shorter ength ofthe debonde! zone, resulting inearler facture. Specimen A, which had 90 debonding butdid have a deep spall, therefore shows more dluctile behavior than Specimen B, which hac a shortdebonded length of only 2 in. ($1 mm), However, Specirmen C,which had adebonded length of § in, (203 mm),eshibited a 38% greater peak displacement than Specimen A. This clearly indicates that thedebonding method, when applied over sulficientlengch, can eféctively delay the fracture ofthe reinforcing bar. tive displacement over a 6.25 in. (159 mm) Computational Analyses Analysis of axially loaded cylindrical prism test For this stu detaecl FE models ofthe three test specimens \were developed using the LS-DYINA’ general-purpose software package. Concrete was modeled using eight node slid ments with mesh size of about 1 in (25 mm, while reinforcing bars were represented by two-node beam elements ig. §) with mesh size of 2 study showed that further refinement of mesh size was not seeded. A continuous surface cap model was used 3s the ‘material model for concrete. The main features of the model include: isotropic constitutive equations, yield surface formulated in terms of a dhree stress invariant shear surface with ‘tanslation for prepeak hardening, a hardening cap that expands and contact, damage based softening with erosion anx! moxlulus reduction, and rate effets for high stain rate applications. Bora! behavior berwcen concrete elements and reinforcing bats was simulated by adding one-dimensional contct allowing movement only along the longitudinal direction based on a predefined bond slip relationship, and preventing penetration normal tothe bar axis Debording betwen reinforcing bats and surrounding concrete was achieved by wetting a very small value of te bond tio of bond stress o slip). Further modeling details in Reference 6, Figute7 shows plots of the applied load versus displacement for the three modeled specimens along with the ploss ofthe previously clescibed experimental imum). A mesh sensitivity Fig. 5: Bond between the reinforcing ber ‘and the eonereto west broken using polyol fi hea-ahink tubing. Het, tubing i shown before ight and ‘flr (of) heat ‘epication Fig. 6: Resullsof loading tede: (@) ‘*pecimen with no bond breaker (Specimen A) ‘exhibited cone: shaped concrete breakout on the tree surface of he crack; ‘and (b) specimens with debonded tinforcing bars (Specimens Band ¢) exhibited no concrete damage on the tree ‘surfice of the “eretck” Displacement, mm 254381 = 0 27 508 Appliee load, kip o os 1% 45 2 25 Displacement, in Fig.7: Load-displacement reaitstor tes specimens (Specimens, B, and C) and corresponding analytical models results. Good agiwement benween computational and experimental results validates the computational method used in this study. Relative tothe fully boreled specimen,an uunborvid lengeh of in. capacity and an unbonded length of § in, (203 mm) has a higher deformation capacity Selected results from the analysis of a specimen with an 8 in, (203 mm) debonded length (Specimen ©) are presented in Fig. 8, The damage index contours indicate mim) has a lower deformation Concrete international DECEMBER 2019 35 i@: © o ° Fig &: Analylical model: (a) schernaiic of frite element model; (b) contour plot of concrete damage index af maximum ‘applied load; and (c) axial foree in reinforcing at maximum ‘applied load, Results shown here ore fora specimen with a dobonded length of 8 in, (203 mm), Specimen C Vertical displacement of central colurmn, mm 0 24 S08 762 t0e «1270524 3896 8 To debonding 5 oor2 g aaa Applied loa6, kip ma Applied load, KN . 0 3 6 50 Vertical displacement of central column, in Fig. 9: Applied vertical laad as a function of vertical dlplacement fora central column loss scenario, Based.on a detailed FE model, ‘hese results venty hat he maximum vettical displacement ofthe Central column increases with the debonded length othe bottom reinforcing bas (on each side of he central column) some level ofconcrete damage starting at abour 8 in. (203 mm} from the gap, corresponding co the location where the concrete was fully bonded to the central winforcing bar. Also, the axial force in the reinfowing bar can be en to be relatively constant within the debongled zone and to gradually decrease with distance away from the debonded zone, Debonded length For this study, FE models of an IMF assembly under a central column loss scenario were evaluated. The design of the modeled specimens was in accordance with ACI 318.02.° Detailed information on the analyses is presented in Reference 6, IMF assemblies with debonded lengths of 0 (no debonding), 1D,20,and 3D were modeled, where D was, the depth of the beam. The models were used to find the 36 oi BER 203 Conerete international ‘optimum debonded length, resulting in the lagestdisplace- ‘mentand peak load prior to bar fracture and beam filure swithout adversely afecting the stiffness ofthe beam. The computational medel was validated with the experimental data forthe case of nodebonding.* Figure 9 depicts the applied vertical load versus the vertical displacement of the central column, based on analyses with differentdebonded lengths. The figure shows that the debonded length of 30 provided the hest formance (largest vertical load and displacemene prior to failure). Because the diffrences in peak loads prior to future ‘were not significant hetween the eases with unbonded Fengths ‘of 2D and 3D. it’s prudent to use 2D as the optimum length of debondi npacting the sine ofthe connection. The results show impronements of more than 309% in the peak load with a debonded length of 2D for the [ME beamcolumn assembly analyzed herein. coavoid adversely i Seismic performance Both the experimental and computational investigations showed that the proposed debongling method can effectively eliminate strain localization of reinforcing bars at wide cracks ‘Computational models showed that debonding of reinforcing bars. the beam eras in a beam-column assembly significantly increases the catenary resistance under acolumn loss

You might also like